DVOL is the “bitcoin VIX”: a single index for how violent a market expects the coming month to be. Several confident claims circulate about it. We took 907 days of our own data (January 2024 — July 2026) and tested three of them — including the one that did not hold up.
Source: the daily DVOL archive for BTC and ETH from Deribit. Every number below is a direct measurement; where something is an estimate or a limitation, we say so plainly. This is a journal of decisions, not signals.
Question 1
Does extreme fear hold, or does it fade?
It fades — always
Question 2
Is a fear peak a price bottom?
Yes, but not a signal
Question 3
Is fear overpriced?
Yes, 74% of the time
Finding 1 · the main one
The most common beginner’s mistake is to see a high DVOL and decide that “it will stay this way now”. The data says the opposite. Over 2.5 years bitcoin DVOL oscillated around a single value — roughly 50 — and it returned there every time, wherever it started from.
In numbers it looks like this: after the calmest days (the lowest 10%) DVOL edged up; after the most anxious ones (the highest 10%, above ~61) it settled by 10 points in a month. The link “high today → high tomorrow” weakened evenly and predictably: over a week, over two, over a month, the memory of the peak melted away.
High volatility falls back to normal. Low volatility does not last either. Extremes do not hold.
Finding 2 · with an honest “but”
The second common belief is more attractive: “buy when everyone is afraid”. We tested it literally — whether the highest fear readings really coincide with a price bottom. Partly, yes.
⚖ Why this is not a buy signal
Here hides the trap most sites keep quiet about. That it was a bottom only becomes known two weeks later — in hindsight. At the moment of a fear peak you do not know whether it is the floor or an intermediate stop before a deeper fall. And most importantly: thirty days after a peak, bitcoin’s return was statistically indistinguishable from any other day. Panic shows where the market broke — but not when to act. We show this coincidence honestly and immediately refuse to sell it as timing.
That is the difference between a journal and a signal service. A signal seller would take the “66%” figure and turn it into a promise. We put a second figure next to it — “after a month it is ≈ 50/50” — and leave you to add them up yourself.
Finding 3 · the cleanest
The third test is the hardest of the three. DVOL shows which moves the market expects. We compared that expectation with what actually happened actually over the following 30 days — and counted how often the fear turned out to be exaggerated.
In plain words: the market almost always pays more for insurance than it eventually turns out to cost. This is neither an anomaly nor our invention — it is a well-known risk premium the literature calls the “fear premium”. But reading about it is one thing; seeing it measured across 877 days of your own data is another. That gap between expectation and fact is the quiet engine behind half the strategies that sell volatility.
The expected move ≈ DVOL ÷ 19 per day. But the market consistently overpays for that expectation.
The limits of the study
A study is honest exactly to the extent that its limits are named honestly. Here are ours:
Two of the three findings (the return to normal and overpriced fear) are solid and published as they are. The third (peak = bottom) we publish only inside this frame: as a description of capitulation, never as a signal.
We look at DVOL every morning — before almost anything else. It answers the question standing in front of every decision: is now the time to buy the move, or is insurance already too expensive? The live value and 2.5 years of history are on the volatility board.
Live DVOL now → The whole Case File of hypotheses